Loan Data From Prosper¶

by Peter Wu¶

Preliminary Wrangling¶

This data set contains 113,937 loans with 81 variables on each loan, including loan amount, borrower rate (or interest rate), current loan status, borrower income, and many others.

# import all packages and set plots to be embedded inline
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sb
%matplotlib inline

df = pd.read_csv('prosperLoanData.csv', encoding='latin-1', low_memory=False)
df.head()

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 113937 entries, 0 to 113936
Data columns (total 81 columns):
ListingKey                             113937 non-null object
ListingNumber                          113937 non-null int64
ListingCreationDate                    113937 non-null object
CreditGrade                            28953 non-null object
Term                                   113937 non-null int64
LoanStatus                             113937 non-null object
ClosedDate                             55089 non-null object
BorrowerAPR                            113912 non-null float64
BorrowerRate                           113937 non-null float64
LenderYield                            113937 non-null float64
EstimatedEffectiveYield                84853 non-null float64
EstimatedLoss                          84853 non-null float64
EstimatedReturn                        84853 non-null float64
ProsperRating (numeric)                84853 non-null float64
ProsperRating (Alpha)                  84853 non-null object
ProsperScore                           84853 non-null float64
ListingCategory (numeric)              113937 non-null int64
BorrowerState                          108422 non-null object
Occupation                             110349 non-null object
EmploymentStatus                       111682 non-null object
EmploymentStatusDuration               106312 non-null float64
IsBorrowerHomeowner                    113937 non-null bool
CurrentlyInGroup                       113937 non-null bool
GroupKey                               13341 non-null object
DateCreditPulled                       113937 non-null object
CreditScoreRangeLower                  113346 non-null float64
CreditScoreRangeUpper                  113346 non-null float64
FirstRecordedCreditLine                113240 non-null object
CurrentCreditLines                     106333 non-null float64
OpenCreditLines                        106333 non-null float64
TotalCreditLinespast7years             113240 non-null float64
OpenRevolvingAccounts                  113937 non-null int64
OpenRevolvingMonthlyPayment            113937 non-null float64
InquiriesLast6Months                   113240 non-null float64
TotalInquiries                         112778 non-null float64
CurrentDelinquencies                   113240 non-null float64
AmountDelinquent                       106315 non-null float64
DelinquenciesLast7Years                112947 non-null float64
PublicRecordsLast10Years               113240 non-null float64
PublicRecordsLast12Months              106333 non-null float64
RevolvingCreditBalance                 106333 non-null float64
BankcardUtilization                    106333 non-null float64
AvailableBankcardCredit                106393 non-null float64
TotalTrades                            106393 non-null float64
TradesNeverDelinquent (percentage)     106393 non-null float64
TradesOpenedLast6Months                106393 non-null float64
DebtToIncomeRatio                      105383 non-null float64
IncomeRange                            113937 non-null object
IncomeVerifiable                       113937 non-null bool
StatedMonthlyIncome                    113937 non-null float64
LoanKey                                113937 non-null object
TotalProsperLoans                      22085 non-null float64
TotalProsperPaymentsBilled             22085 non-null float64
OnTimeProsperPayments                  22085 non-null float64
ProsperPaymentsLessThanOneMonthLate    22085 non-null float64
ProsperPaymentsOneMonthPlusLate        22085 non-null float64
ProsperPrincipalBorrowed               22085 non-null float64
ProsperPrincipalOutstanding            22085 non-null float64
ScorexChangeAtTimeOfListing            18928 non-null float64
LoanCurrentDaysDelinquent              113937 non-null int64
LoanFirstDefaultedCycleNumber          16952 non-null float64
LoanMonthsSinceOrigination             113937 non-null int64
LoanNumber                             113937 non-null int64
LoanOriginalAmount                     113937 non-null int64
LoanOriginationDate                    113937 non-null object
LoanOriginationQuarter                 113937 non-null object
MemberKey                              113937 non-null object
MonthlyLoanPayment                     113937 non-null float64
LP_CustomerPayments                    113937 non-null float64
LP_CustomerPrincipalPayments           113937 non-null float64
LP_InterestandFees                     113937 non-null float64
LP_ServiceFees                         113937 non-null float64
LP_CollectionFees                      113937 non-null float64
LP_GrossPrincipalLoss                  113937 non-null float64
LP_NetPrincipalLoss                    113937 non-null float64
LP_NonPrincipalRecoverypayments        113937 non-null float64
PercentFunded                          113937 non-null float64
Recommendations                        113937 non-null int64
InvestmentFromFriendsCount             113937 non-null int64
InvestmentFromFriendsAmount            113937 non-null float64
Investors                              113937 non-null int64
dtypes: bool(3), float64(50), int64(11), object(17)
memory usage: 68.1+ MB
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 113937 entries, 0 to 113936
Data columns (total 81 columns):
ListingKey                             113937 non-null object
ListingNumber                          113937 non-null int64
ListingCreationDate                    113937 non-null object
CreditGrade                            28953 non-null object
Term                                   113937 non-null int64
LoanStatus                             113937 non-null object
ClosedDate                             55089 non-null object
BorrowerAPR                            113912 non-null float64
BorrowerRate                           113937 non-null float64
LenderYield                            113937 non-null float64
EstimatedEffectiveYield                84853 non-null float64
EstimatedLoss                          84853 non-null float64
EstimatedReturn                        84853 non-null float64
ProsperRating (numeric)                84853 non-null float64
ProsperRating (Alpha)                  84853 non-null object
ProsperScore                           84853 non-null float64
ListingCategory (numeric)              113937 non-null int64
BorrowerState                          108422 non-null object
Occupation                             110349 non-null object
EmploymentStatus                       111682 non-null object
EmploymentStatusDuration               106312 non-null float64
IsBorrowerHomeowner                    113937 non-null bool
CurrentlyInGroup                       113937 non-null bool
GroupKey                               13341 non-null object
DateCreditPulled                       113937 non-null object
CreditScoreRangeLower                  113346 non-null float64
CreditScoreRangeUpper                  113346 non-null float64
FirstRecordedCreditLine                113240 non-null object
CurrentCreditLines                     106333 non-null float64
OpenCreditLines                        106333 non-null float64
TotalCreditLinespast7years             113240 non-null float64
OpenRevolvingAccounts                  113937 non-null int64
OpenRevolvingMonthlyPayment            113937 non-null float64
InquiriesLast6Months                   113240 non-null float64
TotalInquiries                         112778 non-null float64
CurrentDelinquencies                   113240 non-null float64
AmountDelinquent                       106315 non-null float64
DelinquenciesLast7Years                112947 non-null float64
PublicRecordsLast10Years               113240 non-null float64
PublicRecordsLast12Months              106333 non-null float64
RevolvingCreditBalance                 106333 non-null float64
BankcardUtilization                    106333 non-null float64
AvailableBankcardCredit                106393 non-null float64
TotalTrades                            106393 non-null float64
TradesNeverDelinquent (percentage)     106393 non-null float64
TradesOpenedLast6Months                106393 non-null float64
DebtToIncomeRatio                      105383 non-null float64
IncomeRange                            113937 non-null object
IncomeVerifiable                       113937 non-null bool
StatedMonthlyIncome                    113937 non-null float64
LoanKey                                113937 non-null object
TotalProsperLoans                      22085 non-null float64
TotalProsperPaymentsBilled             22085 non-null float64
OnTimeProsperPayments                  22085 non-null float64
ProsperPaymentsLessThanOneMonthLate    22085 non-null float64
ProsperPaymentsOneMonthPlusLate        22085 non-null float64
ProsperPrincipalBorrowed               22085 non-null float64
ProsperPrincipalOutstanding            22085 non-null float64
ScorexChangeAtTimeOfListing            18928 non-null float64
LoanCurrentDaysDelinquent              113937 non-null int64
LoanFirstDefaultedCycleNumber          16952 non-null float64
LoanMonthsSinceOrigination             113937 non-null int64
LoanNumber                             113937 non-null int64
LoanOriginalAmount                     113937 non-null int64
LoanOriginationDate                    113937 non-null object
LoanOriginationQuarter                 113937 non-null object
MemberKey                              113937 non-null object
MonthlyLoanPayment                     113937 non-null float64
LP_CustomerPayments                    113937 non-null float64
LP_CustomerPrincipalPayments           113937 non-null float64
LP_InterestandFees                     113937 non-null float64
LP_ServiceFees                         113937 non-null float64
LP_CollectionFees                      113937 non-null float64
LP_GrossPrincipalLoss                  113937 non-null float64
LP_NetPrincipalLoss                    113937 non-null float64
LP_NonPrincipalRecoverypayments        113937 non-null float64
PercentFunded                          113937 non-null float64
Recommendations                        113937 non-null int64
InvestmentFromFriendsCount             113937 non-null int64
InvestmentFromFriendsAmount            113937 non-null float64
Investors                              113937 non-null int64
dtypes: bool(3), float64(50), int64(11), object(17)
memory usage: 68.1+ MB

df_dict = pd.read_csv('Prosper Loan Data - Variable Definitions - Sheet1.csv')

#display the full rows and columns of the dictionary dataframe
pd.options.display.max_colwidth = 1000
pd.options.display.max_rows = 1000
df_dict

#Copy the dataset, and only select those columns that will be used

cols = ['Term', 'LoanStatus', 'BorrowerAPR', 'BorrowerRate', 'ListingCategory', 'EmploymentStatus'
          , 'IsBorrowerHomeowner', 'CreditScoreRangeLower', 'CreditScoreRangeUpper', 'DebtToIncomeRatioff', 
           'StatedMonthlyIncome', 'PercentFunded', 'Recommendations', 'InvestmentFromFriendsCount', 'ProsperScore']
df_sub = df.copy()
df_sub = df_sub.loc[:, df.columns.isin(cols)]

df_sub

df_sub['LoanStatus'].value_counts()

Current                   56576
Completed                 38074
Chargedoff                11992
Defaulted                  5018
Past Due (1-15 days)        806
Past Due (31-60 days)       363
Past Due (61-90 days)       313
Past Due (91-120 days)      304
Past Due (16-30 days)       265
FinalPaymentInProgress      205
Past Due (>120 days)         16
Cancelled                     5
Name: LoanStatus, dtype: int64

Current                   56576
Completed                 38074
Chargedoff                11992
Defaulted                  5018
Past Due (1-15 days)        806
Past Due (31-60 days)       363
Past Due (61-90 days)       313
Past Due (91-120 days)      304
Past Due (16-30 days)       265
FinalPaymentInProgress      205
Past Due (>120 days)         16
Cancelled                     5
Name: LoanStatus, dtype: int64

# rename the LoanStatus column for better visualization, and make it categorical
def renameLoanStatus(x):
    if 'Past Due' in x:
        return 'Past Due'
    elif x == 'FinalPaymentInProgress':
        return 'InProgress'
    return x
        
df_sub.LoanStatus = df_sub.LoanStatus.apply(renameLoanStatus);
df_sub.LoanStatus = df_sub.LoanStatus.astype('category')

df_sub.EmploymentStatus.value_counts()

Employed         67322
Full-time        26355
Self-employed     6134
Not available     5347
Other             3806
Part-time         1088
Not employed       835
Retired            795
Name: EmploymentStatus, dtype: int64

Employed         67322
Full-time        26355
Self-employed     6134
Not available     5347
Other             3806
Part-time         1088
Not employed       835
Retired            795
Name: EmploymentStatus, dtype: int64

#  make EmploymentStatus categorical
df_sub.EmploymentStatus = df_sub.EmploymentStatus.astype('category')

# overview of the datatypes and infomation
df_sub.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 113937 entries, 0 to 113936
Data columns (total 13 columns):
Term                          113937 non-null int64
LoanStatus                    113937 non-null category
BorrowerAPR                   113912 non-null float64
BorrowerRate                  113937 non-null float64
ProsperScore                  84853 non-null float64
EmploymentStatus              111682 non-null category
IsBorrowerHomeowner           113937 non-null bool
CreditScoreRangeLower         113346 non-null float64
CreditScoreRangeUpper         113346 non-null float64
StatedMonthlyIncome           113937 non-null float64
PercentFunded                 113937 non-null float64
Recommendations               113937 non-null int64
InvestmentFromFriendsCount    113937 non-null int64
dtypes: bool(1), category(2), float64(7), int64(3)
memory usage: 9.0 MB
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 113937 entries, 0 to 113936
Data columns (total 13 columns):
Term                          113937 non-null int64
LoanStatus                    113937 non-null category
BorrowerAPR                   113912 non-null float64
BorrowerRate                  113937 non-null float64
ProsperScore                  84853 non-null float64
EmploymentStatus              111682 non-null category
IsBorrowerHomeowner           113937 non-null bool
CreditScoreRangeLower         113346 non-null float64
CreditScoreRangeUpper         113346 non-null float64
StatedMonthlyIncome           113937 non-null float64
PercentFunded                 113937 non-null float64
Recommendations               113937 non-null int64
InvestmentFromFriendsCount    113937 non-null int64
dtypes: bool(1), category(2), float64(7), int64(3)
memory usage: 9.0 MB

# descriptive statistics for numeric variables
df_sub.describe()

There are 113937 loans within the dataset, and initially with 81 variables on each loan. Some variables are just IDs of the loans, while others are personal infomation about the borrowers. I only extracted a fraction of the variables to analyze on, which I think will be intersting to work with. There are 14 columns I used:¶

Term                      
LoanStatus                    
BorrowerAPR                   
BorrowerRate                                   
EmploymentStatus              
IsBorrowerHomeowner          
CreditScoreRangeLower         
CreditScoreRangeUpper         
StatedMonthlyIncome           
PercentFunded                 
Recommendations               
InvestmentFromFriendsCount
ProsperScore

Most variables are numertic, while there are 3 categorical ones.

I am interested in two things: first, which features are best at predicting the `LoanStatus`; second, which features determine the `BorrowerAPR` and `BorrowerRate` at most?¶

I expect the two `CreditScoreRange` will have the most effect on loan completion, and variables like `Occupation`, `EmploymentStatus`, `IsBorrowerHomeowner` will determine the borrowers's interest rate.¶

Univariate Exploration¶

I'll start by looking at the loan status.

# absolute Frequency of LoanStatus
base_color = sb.color_palette()[4]
plt.figure(figsize=(10,5))
sb.countplot(data=df_sub, x='LoanStatus', color=base_color);

# get the relative frequency of LoanStatus, i.e., by proportion
n_points = df_sub.shape[0]
max_count = df_sub['LoanStatus'].value_counts().max()
max_prop = max_count / n_points

# generate tick mark locations and names
tick_props = np.arange(0, max_prop, 0.05)
tick_names = ['{:0.2f}'.format(v) for v in tick_props]

# create the plot
plt.figure(figsize=(10,5))
sb.countplot(data = df_sub, x = 'LoanStatus', color = base_color, )
plt.yticks(tick_props * n_points, tick_names)
plt.ylabel('proportion')

locs, labels = plt.xticks() # get the current tick locations and labels

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = df_sub['LoanStatus'].value_counts()[label.get_text()]
    pct_string = '{:0.3f}%'.format(100*count/n_points)

    # print the annotation just below the top of the bar
    plt.text(loc, count-8, pct_string, ha = 'center', color = 'black')

plt.xticks(rotation = 15);

About half of the loans are ongoing, and with more than 5% of the loans are bad (means they are either defaulted or past due).

Next up, the Term variable.

It would be better to categorize Term.

df_sub['Term'] = df_sub['Term'].astype('category')

# get the relative frequency of Term, i.e., by proportion
n_points = df_sub.shape[0]
max_count = df_sub['Term'].value_counts().max()
max_prop = max_count / n_points

# generate tick mark locations and names
tick_props = np.arange(0, max_prop, 0.05)
tick_names = ['{:0.2f}'.format(v) for v in tick_props]

# create the plot
plt.figure(figsize=(10,5))
sb.countplot(data = df_sub, x = 'Term', color = base_color, )
plt.yticks(tick_props * n_points, tick_names)
plt.ylabel('proportion')

locs, labels = plt.xticks() # get the current tick locations and labels

# also loops over the value_counts series
j = 0

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = df_sub['Term'].value_counts()[j]
    pct_string = '{:0.3f}%'.format(100*count/n_points)

    # print the annotation just below the top of the bar
    plt.text(loc, count-8, pct_string, ha = 'center', color = 'black')

    j += 1

The majority of of term is 36 months.

Let's take a look at the EmploymentStatus and IsBorrowerHomeowner, the two categorical variables

# get the relative frequency of EmploymentStatus, i.e., by proportion
n_points = df_sub.shape[0]
max_count = df_sub['EmploymentStatus'].value_counts().max()
max_prop = max_count / n_points

# generate tick mark locations and names
tick_props = np.arange(0, max_prop, 0.05)
tick_names = ['{:0.2f}'.format(v) for v in tick_props]

# create the plot
plt.figure(figsize=(10,5))
sb.countplot(data = df_sub, x = 'EmploymentStatus', color = base_color, )
plt.yticks(tick_props * n_points, tick_names)
plt.ylabel('proportion')

locs, labels = plt.xticks() # get the current tick locations and labels

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = df_sub['EmploymentStatus'].value_counts()[label.get_text()]
    pct_string = '{:0.3f}%'.format(100*count/n_points)

    # print the annotation just below the top of the bar
    plt.text(loc, count-8, pct_string, ha = 'center', color = 'black')

plt.xticks(rotation = 15);

# do the same to IsBorrowerHomeowner, but create a pie chart instead
n_points = df_sub.shape[0]
max_count = df_sub['IsBorrowerHomeowner'].value_counts().max()
max_prop = max_count / n_points

# generate tick mark locations and names
tick_props = np.arange(0, max_prop, 0.05)
tick_names = ['{:0.2f}'.format(v) for v in tick_props]

# create the plot
plt.figure(figsize=(10,5))
sb.countplot(data = df_sub, x = 'IsBorrowerHomeowner', color = base_color, )
plt.yticks(tick_props * n_points, tick_names)
plt.ylabel('proportion')

locs, labels = plt.xticks() # get the current tick locations and labels

# loop through each pair of locations and labels

# also loop over the value_counts series
j = 0

for loc, label in zip(locs, labels):

    # get the index of series to get the boolean variable
    count = df_sub['IsBorrowerHomeowner'].value_counts()[j]
    pct_string = '{:0.3f}%'.format(100*count/n_points)

    # print the annotation just below the top of the bar
    plt.text(loc, count-8, pct_string, ha = 'center', color = 'black')
    
    j += 1

The majority of th borrowers are employed, however, wouldn't the 'Employed' column being more general than the 'Full-time' and 'Part-time'?

Approximately half of the borrowers owned a house.

# CreditScoreRangeLower

plt.hist(data=df_sub, x='CreditScoreRangeLower');

# noting that there are some outliers below the 200 ticks

/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:824: RuntimeWarning: invalid value encountered in greater_equal
  keep = (tmp_a >= first_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:825: RuntimeWarning: invalid value encountered in less_equal
  keep &= (tmp_a <= last_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:824: RuntimeWarning: invalid value encountered in greater_equal
  keep = (tmp_a >= first_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:825: RuntimeWarning: invalid value encountered in less_equal
  keep &= (tmp_a <= last_edge)

# examine those data with a very low credit score
df_sub.query('CreditScoreRangeLower < 200')

All of these rows have a credit score lower bound of 0, and upper bound of 19.0. Moreover, many of them were either defaulted or charged off.

df_sub.query('CreditScoreRangeLower < 200').LoanStatus.value_counts()

Defaulted     59
Completed     39
Chargedoff    35
Past Due       0
InProgress     0
Current        0
Cancelled      0
Name: LoanStatus, dtype: int64

Defaulted     59
Completed     39
Chargedoff    35
Past Due       0
InProgress     0
Current        0
Cancelled      0
Name: LoanStatus, dtype: int64

As we can see, more than half of the debts went bad. These outliers should be dropped, otherwise the analysis twowards credit score would become difficult.

df_sub.drop(axis=0, index=df_sub.query('CreditScoreRangeLower < 200').index, inplace=True)

print(df_sub.CreditScoreRangeLower.describe())
print(df_sub.CreditScoreRangeUpper.describe())

count    113213.000000
mean        686.373120
std          62.201999
min         360.000000
25%         660.000000
50%         680.000000
75%         720.000000
max         880.000000
Name: CreditScoreRangeLower, dtype: float64
count    113213.000000
mean        705.373120
std          62.201999
min         379.000000
25%         679.000000
50%         699.000000
75%         739.000000
max         899.000000
Name: CreditScoreRangeUpper, dtype: float64
count    113213.000000
mean        686.373120
std          62.201999
min         360.000000
25%         660.000000
50%         680.000000
75%         720.000000
max         880.000000
Name: CreditScoreRangeLower, dtype: float64
count    113213.000000
mean        705.373120
std          62.201999
min         379.000000
25%         679.000000
50%         699.000000
75%         739.000000
max         899.000000
Name: CreditScoreRangeUpper, dtype: float64

# take a look at CreditScorelower and upper again.

plt.figure(figsize=(10, 5))

plt.subplot(1,2,1)
bins_edge = np.arange(df_sub.CreditScoreRangeLower.min(), df_sub.CreditScoreRangeLower.max()+10, 10)
plt.hist(data=df_sub, x='CreditScoreRangeLower', bins=bins_edge)
plt.xlabel('CreditScoreRangeLower')
plt.ylabel('Counts');

plt.subplots_adjust(wspace= 0.3)

plt.subplot(1,2,2)
bins_edge = np.arange(df_sub.CreditScoreRangeUpper.min(), df_sub.CreditScoreRangeUpper.max()+10, 10)
plt.hist(data=df_sub, x='CreditScoreRangeUpper', bins=bins_edge)
plt.xlabel('CreditScoreRangeUpper')
plt.ylabel('Counts');

The two bounds of credit score is very similar, noting there are gaps between every sub-interval of the graph, this might indicate the credit scores were given with a sepcific range for each.

The credit socres look very normally distributed as well.

# combine the two plots

bins_edge = np.arange(df_sub.CreditScoreRangeLower.min(), df_sub.CreditScoreRangeUpper.max()+10, 10)
plt.hist(data=df_sub, x='CreditScoreRangeLower', bins=bins_edge)

plt.hist(data=df_sub, x='CreditScoreRangeUpper', bins=bins_edge)
plt.xlabel('CreditScoreRangeCombined')
plt.ylabel('Counts')
plt.legend(['CreditScoreRangeLower', 'CreditScoreRangeUpper']);

Interestingly, thw two graphs fill in each other's gap perfectly, this further indicates the lower and upper bound of credit scores were givien with some certain values.

# take a look at the StatedMonthlyIncome variable
df_sub.StatedMonthlyIncome.describe()

count    1.138040e+05
mean     5.611754e+03
std      7.481310e+03
min      0.000000e+00
25%      3.208333e+03
50%      4.666667e+03
75%      6.833333e+03
max      1.750003e+06
Name: StatedMonthlyIncome, dtype: float64

count    1.138040e+05
mean     5.611754e+03
std      7.481310e+03
min      0.000000e+00
25%      3.208333e+03
50%      4.666667e+03
75%      6.833333e+03
max      1.750003e+06
Name: StatedMonthlyIncome, dtype: float64

print(df_sub.query('StatedMonthlyIncome == 0')['Term'].count())
print('{:.5%}'.format(df_sub.query('StatedMonthlyIncome == 0')['Term'].count() / df_sub['Term'].count()))

1393
1.22403%
1393
1.22403%

There are 1393 borrowers with a monthly income of 0, however, it is not significant because it acccounts for only 1.22403% in porpotion of the total.

print(df_sub.query('StatedMonthlyIncome >= 50000')['Term'].count())
print('{:.5%}'.format(df_sub.query('StatedMonthlyIncome >= 50000')['Term'].count() / df_sub['Term'].count()))

91
0.07996%
91
0.07996%

There are 18 borrowers with a monthly income higher than 100000. Altough they might be wrong values, their existence make plotting difficult in the later section. They should better be dropped.

log_binsize = 0.025
bins = 10 ** np.arange(0.0, np.log10(df_sub.StatedMonthlyIncome.max())+log_binsize, log_binsize)

# insert a bin at 0, since there are some values = 0 for StatedMonthlyIncome
bins = list(bins)
bins.insert(0, 0)

plt.figure(figsize= (10, 5))
plt.hist(data=df_sub, x='StatedMonthlyIncome', bins=bins)

# change the graph to log scale
plt.xscale('log');
plt.xticks([500, 1e3, 2e3, 5e3, 1e4, 2e4, 5e4], [ 500, '1k', '2k', '5k', '10k', '20k', '50k'])

plt.xlabel('StatedMonthlyIncome ($)')
plt.ylabel('Count');

The majorty of values are normally distributed, however, outliers are significant at around 0 while also some greater than 50k.

df_sub.query('StatedMonthlyIncome == 0')

These rows seemed to be wrong in the StatedMonthlyIncome values. Many of them are employed of even full-time employed, and how would they have a StatedMonthlyIncome of 0? In addition, many of their credits are above average (>= 650 approximately), it would be hard to have a high credit score without a stable monthly income. Therefore, I would drop all of these rows, as considered them are outliers.

df_sub.drop(index=df_sub.query('StatedMonthlyIncome == 0').index, inplace = True);

df_sub.query('StatedMonthlyIncome >= 50000')

These rows seem not to be wrong, as the variables all look valid. Nevertheless, just as stated above, their existence affect the drawing of correlation graphs in the later section, it is not ideal to plot with such great outliers. Hence all StatedMonthlyIncome >= 50000 can be dropped, they are really too far from majority of values.

df_sub.drop(index=df_sub.query('StatedMonthlyIncome >= 50000').index, inplace = True);

# plot the StatedMonthlyIncome again
log_binsize = 0.025
bins = 10 ** np.arange(0.0, np.log10(df_sub.StatedMonthlyIncome.max())+log_binsize, log_binsize)

# insert a bin at 0, since there are some values = 0 for StatedMonthlyIncome
bins = list(bins)
bins.insert(0, 0)

plt.figure(figsize= (10, 5))
plt.hist(data=df_sub, x='StatedMonthlyIncome', bins=bins)

# change the graph to log scale
plt.xscale('log');
plt.xticks([500, 1e3, 2e3, 5e3, 1e4, 2e4, 5e4], [ 500, '1k', '2k', '5k', '10k', '20k', '50k'])

plt.xlabel('StatedMonthlyIncome ($)')
plt.ylabel('Count');

There is still a group of outliers around 0, but fewer than before, take a look at them.

df_sub.query('StatedMonthlyIncome <= 200')

Many of these incomes are problematic, values like: 0.083333, 3.333333, 9.416667 2.500000 appeared in these rows. Does these look like realistic incomes? Especially for 0.083333, which showed up multiple times in these rows, it is more possible to be a input error rather than real values. Therefore, all of these rows could be dropped, as they are either outliers or wrong values.

df_sub.drop(index=df_sub.query('StatedMonthlyIncome <= 200').index, inplace=True)

# plot the StatedMonthlyIncome again
log_binsize = 0.025
bins = 10 ** np.arange(0.0, np.log10(df_sub.StatedMonthlyIncome.max())+log_binsize, log_binsize)

# insert a bin at 0, since there are some values = 0 for StatedMonthlyIncome
bins = list(bins)
bins.insert(0, 0)

plt.figure(figsize= (10, 5))
plt.hist(data=df_sub, x='StatedMonthlyIncome', bins=bins)

# change the graph to log scale
plt.xscale('log');
plt.xticks([500, 1e3, 2e3, 5e3, 1e4, 2e4, 5e4], [ 500, '1k', '2k', '5k', '10k', '20k', '50k'])

plt.xlabel('StatedMonthlyIncome ($)')
plt.ylabel('Count');

This time the graph looks better in distribution with no visible outliers.

Take a look at the PercentFunded variable, it refers to the amound of loan that were funded to a borrower.

# plot the distribution of PercentFunded
bins_edge =  np.arange(0, df_sub.PercentFunded.max() + 0.01, 0.01)
plt.hist(data=df_sub, x='PercentFunded', bins=bins_edge);

print(df_sub.query('0.99 <= PercentFunded <= 1.01')['Term'].count())
print('{:.5%}'.format(df_sub.query('0.99 <= PercentFunded <= 1.01')['Term'].count() / df_sub['Term'].count()))

110995
99.26309%
110995
99.26309%

The vast majority of PercentFunded is in the interval between 0.99 and 1.01. The variable is so concentrated and would less of interest in later analysis.

Take a look at the Recommendations variable.

bins_edge = np.arange(0, df_sub.Recommendations.max()+1, 1)
plt.hist(data=df_sub, x='Recommendations', bins=bins_edge);

print(df_sub.query('Recommendations == 0')['Term'].count())
print('{:.5%}'.format(df_sub.query('Recommendations == 0')['Term'].count() / df_sub['Term'].count()))

107778
96.38612%
107778
96.38612%

More than 96% of the Recommendations are 0, again, the variable would be less of interest in further analysis.

bins_edge = np.arange(0, df_sub.InvestmentFromFriendsCount.max()+1, 1)
plt.hist(data=df_sub, x='InvestmentFromFriendsCount', bins=bins_edge);

print(df_sub.query('InvestmentFromFriendsCount == 0')['Term'].count())
print('{:.5%}'.format(df_sub.query('InvestmentFromFriendsCount == 0')['Term'].count() / df_sub['Term'].count()))

109843
98.23286%
109843
98.23286%

Similarly, 98% of InvestmentFromFriendsCount is 0, which makes the variable less of further interest.

Next up, ProsperScore.

It would be better to categorize ProsperScore.

df_sub['ProsperScore'] = df_sub['ProsperScore'].astype('category')

# get the relative frequency of ProsperScore, i.e., by proportion
n_points = df_sub.shape[0]
max_count = df_sub['ProsperScore'].value_counts().max()
max_prop = max_count / n_points

# generate tick mark locations and names
tick_props = np.arange(0, max_prop, 0.05)
tick_names = ['{:0.2f}'.format(v) for v in tick_props]

# create the plot
plt.figure(figsize=(10,5))
sb.countplot(data = df_sub, x = 'ProsperScore', color = base_color, )
plt.yticks(tick_props * n_points, tick_names)
plt.ylabel('proportion')

locs, labels = plt.xticks() # get the current tick locations and labels

# also loops over the value_counts series
j = 1

# loop through each pair of locations and labels
for loc, label in zip(locs, labels):

    # get the text property for the label to get the correct count
    count = df_sub['ProsperScore'].value_counts()[j]
    pct_string = '{:0.3f}%'.format(100*count/n_points)

    # print the annotation just below the top of the bar
    plt.text(loc, count+100, pct_string, ha = 'center', color = 'black')

    j += 1

Finally, BorrowerAPR and BorrowerRate are anlayzed, they would be the 'to-indicators' like LoanStatus (means using other variables to determine them).

df_sub.head(10)

# plot the distributions of BorrowerAPR and BorrowerRate
plt.figure(figsize=(8,4))

bins_edge = np.arange(0, df_sub.BorrowerAPR.max()+0.002, 0.002)
plt.subplot(1,2,1)
plt.hist(data=df_sub, x='BorrowerAPR', bins=bins_edge)
plt.xlabel('BorrowerAPR')
plt.ylabel('count')

plt.subplots_adjust(wspace=0.4)

bins_edge = np.arange(0, df_sub.BorrowerRate.max()+0.002, 0.002)
plt.subplot(1,2,2)
plt.hist(data=df_sub, x='BorrowerRate', bins=bins_edge)
plt.xlabel('BorrowerRate')
plt.ylabel('count');

The two distributions look very similar, and they have a extremly high count at around 0.35 and 0.3 respectively.

# combine the two plots
plt.figure(figsize=(8,6))
bins_edge = np.arange(0, df_sub.BorrowerAPR.max()+0.002, 0.002)
plt.hist(data=df_sub, x='BorrowerAPR', bins=bins_edge)

bins_edge = np.arange(0, df_sub.BorrowerRate.max()+0.002, 0.002)
plt.hist(data=df_sub, x='BorrowerRate', bins=bins_edge)
plt.xlabel('BorrowerRate/APR combined')
plt.ylabel('count')
plt.legend(['BorrowerAPR', 'BorrowerRate']);

Not surprisingly, the two graphs look very alike.

The `StatedMonthlyIncome` took a very long range of values, as such a log transform was applied. After the transformation ,the variable looks to be normally distributed, but still with a extremely long tail in both directions.¶

During the above process, many invalid income values were found, and were excluded from the dataset, a significant amount of outliers were thus removed. This is important for bivariate and multivariate analysis in the later part, because those values would enormously drag the average income low.¶

Briefly, after the univariate exploration, three variables were deemed to be less value of analysis, and would not be used in the next part. They are:¶

`PercentFunded`, `Recommendations`, `InvestmentFromFriendsCount`.¶

The rest will be used to move forwards are:¶

`Term`, `LoanStatus`, `BorrowerAPR`, `BorrowerRate`, `EmploymentStatus`, `IsBorrowerHomeowner`,¶

`CreditScoreRangeLower`, `ProsperScore`, `CreditScoreRangeUpper`, and `StatedMonthlyIncome` (10 variables in total).¶

Bivariate Exploration¶

To start off with, I want to look at the pairwise correlations present between features in the data.

This part will focus on the relationship between: first, `BorrowerAPR` and `BorrowerRate` and the other variables, to dig out the¶

determinants of brrowers' interest rate; second, `LoanStatus`and the other variables, to dig out what would be the potential reason for a loan to go bad.¶

I will start with `BorrowerAPR` and `BorrowerRate` (`Borrowers rate`).¶

numeric_vars = ['BorrowerAPR', 'BorrowerRate', 'CreditScoreRangeLower', 'CreditScoreRangeUpper', 
                'StatedMonthlyIncome']
categoric_vars = ['Term', 'LoanStatus', 'EmploymentStatus', 'IsBorrowerHomeowner', 'ProsperScore']

# correlation plot
plt.figure(figsize = [14, 7])
sb.heatmap(df_sub[numeric_vars].corr(), annot = True, fmt = '.3f',
           cmap = 'Blues_r', center = 0.25)
plt.show()

# plot matrix: sample 500 loans so that plots are clearer and
# they render faster
samples = np.random.choice(df_sub.shape[0], 500, replace = False)
df_sub_samp = df_sub.loc[samples,:]

g = sb.PairGrid(data = df_sub_samp, vars = numeric_vars)
g = g.map_diag(plt.hist, bins = 50);
g.map_offdiag(plt.scatter, color='Grey');

/anaconda3/lib/python3.7/site-packages/pandas/core/indexing.py:1494: FutureWarning: 
Passing list-likes to .loc or [] with any missing label will raise
KeyError in the future, you can use .reindex() as an alternative.

See the documentation here:
https://pandas.pydata.org/pandas-docs/stable/indexing.html#deprecate-loc-reindex-listlike
  return self._getitem_tuple(key)
/anaconda3/lib/python3.7/site-packages/pandas/core/indexing.py:1494: FutureWarning: 
Passing list-likes to .loc or [] with any missing label will raise
KeyError in the future, you can use .reindex() as an alternative.

See the documentation here:
https://pandas.pydata.org/pandas-docs/stable/indexing.html#deprecate-loc-reindex-listlike
  return self._getitem_tuple(key)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:824: RuntimeWarning: invalid value encountered in greater_equal
  keep = (tmp_a >= first_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:825: RuntimeWarning: invalid value encountered in less_equal
  keep &= (tmp_a <= last_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:824: RuntimeWarning: invalid value encountered in greater_equal
  keep = (tmp_a >= first_edge)
/anaconda3/lib/python3.7/site-packages/numpy/lib/histograms.py:825: RuntimeWarning: invalid value encountered in less_equal
  keep &= (tmp_a <= last_edge)

As observed from above, there is a moderate negative raltionship between the two credit score bounds and the two borrorwers rate. Meanwhile, the two pairs of vriables are almost podtively related to each other perfectly. The rest numeric variabels are not as related to the two borrorwers rate.

Let's see how borrowers rate interact with other categorical variables.

First off, see how ProsperScore, Term, EmploymentStatus, and IsBorrowerHomeowner relates to borrowers rate.

# plot matrix of Borrowers rate against categorical features.
i = 1
def scattergrid(x, y, **kwargs):
    """ Quick hack for creating scatter plots with pyplot's subplot. """
    
    
    global i
    plt.subplot(4,2,i)
    plt.subplots_adjust(wspace=0.2, hspace= 1, )
    sb.stripplot(x, y, color = base_color, alpha = 0.1)
    i += 1

plt.figure(figsize=(20, 20))
scattergrid(df_sub.ProsperScore, df_sub.BorrowerAPR)
scattergrid(df_sub.ProsperScore, df_sub.BorrowerRate)
scattergrid(df_sub.Term, df_sub.BorrowerAPR)
scattergrid(df_sub.Term, df_sub.BorrowerRate)
i = 1

i = 1
plt.figure(figsize=(20, 20))
scattergrid(df_sub.EmploymentStatus, df_sub.BorrowerAPR)
plt.xticks(rotation = -10)
scattergrid(df_sub.EmploymentStatus, df_sub.BorrowerRate)
plt.xticks(rotation = -10)
scattergrid(df_sub.IsBorrowerHomeowner, df_sub.BorrowerAPR)
scattergrid(df_sub.IsBorrowerHomeowner, df_sub.BorrowerRate);

As observed from above, ProsperScore seemed to be negatively related to the two borrowers rate. When ProsperScore was as low as 1.0, typical borrowers rate went up high to 3.0. In the middle, however, Borrowers rate could take a wide range of vlaue, from 0.1 to 0.3.

The other variables like Term, EmploymentStatus, and IsBorrowerHomeowner, do not have a visible correlation with Borrowers rate.

The disciussion will now shift from `BorrowerAPR` and `BorrowerRate` to `LoanStatus`, to see how other variables interact with it.¶

# plot matrix of LoanStatus against numeric features.
i = 1
def Boxgrid(x, y, **kwargs):
    """ Quick hack for creating scatter plots with pyplot's subpplot """
    
    
    global i
    plt.subplot(2,2,i)
    plt.subplots_adjust(wspace=0.2, hspace= 0.3, )
    sb.violinplot(x, y, color = base_color)
    i += 1

plt.figure(figsize=(15, 15))
Boxgrid(df_sub.LoanStatus, df_sub.CreditScoreRangeLower)
Boxgrid(df_sub.LoanStatus, df_sub.CreditScoreRangeUpper)
Boxgrid(df_sub.LoanStatus, df_sub.StatedMonthlyIncome)
i = 1

Just as expected, the CreditScore Bounds has wave-like distribution, this is most visible in the Current LoanStatus. And not suprisingly, the Defaulted, Cancelled, Chargedoff and Past Due LoanStatus, which all are bad loans, have a lower median CreditScore Bounds than the Current and InProgress LoanStatus, which are good loans (or about to be good loans). This might reveal that good debts came from a high credit score borrower on average.

Despite the extremely long tails (indicates great outliers) in StatedMonthlyIncome, it is still visible that the Current and InProgress LoanStatus have a higher income in medians.

# plot matrix of LoanStatus against categorical features.

i = 1
def Countgrid(x, hue, subtitile, **kwargs):
    """ Quick hack for creating counts plots with pyplot's subplot. """
    
    
    global i
    plt.subplot(2,2,i)
    plt.subplots_adjust(wspace=0.2, hspace= 0.2, )
    sb.countplot(x, hue=hue, palette='Reds')
    plt.legend(loc=1)
    plt.gca().set_title(subtitile)
    i += 1
    
plt.figure(figsize=(15, 10))
Countgrid(df_sub.LoanStatus, df_sub.Term, "Term")
Countgrid(df_sub.LoanStatus, df_sub.EmploymentStatus, 'EmploymentStatus')
Countgrid(df_sub.LoanStatus, df_sub.IsBorrowerHomeowner, 'IsBorrowerHomeowner')
Countgrid(df_sub.LoanStatus, df_sub.ProsperScore, 'ProsperScore')
i = 1

60 months' Term seem to have a better performance in LoanStatus, compared to the mode 36 months, fewer loans were Chargedoff or went Defaulted in 60 months loans.

EmploymentStatus did not give much suprise, as the plot is relatively constant on each column.

Is borrowers did not own a home (IsBorrowerHomeowner), they had a sligtly higher chance of making the loans bad. However, the differece should not be statistically significant enough.

ProsperScore showed that borrowers with a higher score are less likely in making the loans bad.

Finally, let's examine the relationships between the three pivot vriables, LoanStatus, BorrowerAPR, and BorrowerRate.

# plot matrix of LoanStatus against numeric features.
i = 1
def Boxgrid(x, y, **kwargs):
    """ Quick hack for creating point plots with seaborn's pointplot """
    
    
    global i
    plt.subplot(1,2,i)
    plt.subplots_adjust(wspace=0.2, hspace= 0.3, )
    sb.pointplot(x, y, linestyles = ":", color=base_color)
    plt.title('Avg. value')
    i += 1

plt.figure(figsize=(17, 5))
Boxgrid(df_sub.LoanStatus, df_sub.BorrowerAPR)
Boxgrid(df_sub.LoanStatus, df_sub.BorrowerRate)
i = 1

The general trend is evident, Current and Completed LoanStatus has the lowest average borrowers rate overall, if we considered the great errorlines in Cancelled. Chargedoff, Defaulted, and Past Due all have high borrowers rate on average.

For `BorrowerAPR` and `BorrowerRate`: the pair of `CreditScore Bounds` and `ProsperScore` seemed to be both negatively related to them.¶

For `LoanStatus`: `CreditScore Bounds`, `Term`, `ProsperScore` seemed to affect it in various ways.¶

Interestingly, the two pairs of variables: `BorrowerAPR` and `BorrowerRate`, `CreditScoreRangeLower` and `CreditScoreRangeUpper` are almost perfectly related to each other.¶

There is also an interplay betweeen the `three pivot variables`, although their relationships were not intended to be analyzed initially.¶

#### Given the extreme similarity in distributions between `BorrowerAPR` and `BorrowerRate` , and as well as between `CreditScoreRangeLower` and `CreditScoreRangeUpper`, I would only use one of each pair to analyze in the next section: namely, `BorrowerAPR` and `CreditScoreRangeLower`. ####

Multivariate Exploration¶

In this section, I will analyze the relationships between each 3 of the variables, to see how they affect the pivot variables in a deeper level of details.

I will start by looking at BorrowerAPR, CreditScoreRangeLower, and ProsperScore.

# create faceted heat maps on levels of the ProsperScore variable
bins_x = np.arange(df_sub.CreditScoreRangeLower.min()+150, df_sub.CreditScoreRangeLower.max(), 25.0)
bins_y = np.arange(0.0, 0.3, 0.05)

i = 1
def hist2dgrid(ProsperScore, **kwargs):
    """ Quick hack for creating heatmap plots with pyplot's hist2d """
    
    df_sub_temp = df_sub[df_sub.ProsperScore == ProsperScore]
    
    global i
    plt.subplot(6,2,i)
    plt.subplots_adjust(wspace=0.1, hspace= 0.5, )
    plt.hist2d(data=df_sub_temp, x='CreditScoreRangeLower', y='BorrowerRate' , cmin=0.5, 
               cmap = 'viridis_r', bins = [bins_x, bins_y])
    plt.gca().set_title('ProsperScore = '+ str(i))
    i = int(i)
    i += 1

plt.figure(figsize=(14,17))
for i in range(1, 12):
    hist2dgrid(i)

plt.colorbar()
plt.show()

The trend is most obvious in high ProsperScore, like 9.0, 10.0, and 11.0. In high ProsperScore, the interst rate decreases as borrowers' CreditScoreRangeLower increases. However, the trend is not so evident in low ProsperScore.

Next up, I will move on to LoanStatus, and starting with its relationship with Term and CreditScoreRangeLower.

# plot the matrices between LoanStatus, Term, and CreditScoreRangeLower. 
g = sb.FacetGrid(data = df_sub, col = 'LoanStatus', height = 4, col_wrap=3)
g.map(sb.boxenplot, 'Term', 'CreditScoreRangeLower', color=base_color,);

/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the boxenplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)
/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the boxenplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)

Let's look at each plot individually, the Cancelled plot can not provide much information since it only has the Term of 36; in the Chargedoff plot, lower median of CreditScore is observed in Term 36; in btoh Completed and Current plots, differnt Term values did not show much difference in median; in Defaulted and Inprogress plots, the Term36 revealed a low median Creditscore; finally, the Past Due plot did show significant enough difference in medians.

Briefly, if a loan was borrowed in a Term of 36, then it is more likely to go bad, if the borrower's CreditScore is below 700.

Now, we can move on to the pair LoanStaus, Term, and ProsperScore.

# plot the matrices between LoanStatus, Term, and CreditScoreRangeLower. 
g = sb.FacetGrid(data = df_sub, col = 'LoanStatus', height = 4, col_wrap=3)
g.map(sb.boxenplot, 'ProsperScore', 'CreditScoreRangeLower', color=sb.color_palette()[6]);

/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the boxenplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)
/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the boxenplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)

One interesting fact from the plots: if a borrower has a high ProsperScore of 11.0 and a high CreditScoreRangeLower, then his/her loan is almost guaranteed to not go bad. This is witnessed in the Defaulted, Chargedoff, and PastDue boxplots, they have close to zero amount of data with high ProsperScore and high CreditScoreRangeLower at the same time.

Finally, we will look the three categoricals: LoanStatus, Term, and ProsperScore.

# plot the matrices between LoanStatus, Term, and ProsperScore. 

#since our attention can be focused on the bad loans distribution, I will limit the LoanSatus here.
LoanStatus_sub = ['Cancelled', 'Chargedoff', 'Defaulted', 'Past Due']
df_sub_temp = df_sub.copy()
df_sub_temp  = df_sub_temp[df_sub_temp['LoanStatus'].isin(LoanStatus_sub)]
LoanStatus_sub_order = pd.api.types.CategoricalDtype(categories=LoanStatus_sub, ordered=True)
df_sub_temp.LoanStatus =  df_sub_temp.LoanStatus.astype(LoanStatus_sub_order)

g = sb.FacetGrid(data = df_sub_temp, col = 'Term',row='ProsperScore', height = 4, aspect=1.5)
g.map(sb.countplot, 'LoanStatus', color=sb.color_palette()[7]);

/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the countplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)
/anaconda3/lib/python3.7/site-packages/seaborn/axisgrid.py:715: UserWarning: Using the countplot function without specifying `order` is likely to produce an incorrect plot.
  warnings.warn(warning)

Just as expected, median ProsperScore like 5.0, 6.0 and 7.0, and the Term of 36 contributed most to bad loans. The low ProsperScore though, might have fewer data in total, and therefore resulted in less count of bad loans.

A great deal of interplays were obsereved in this session:¶

In high ProsperScore, the interst rate decreases as borrowers' CreditScoreRangeLower increases.
If a loan was borrowed in a Term of 36, then it is more likely to go bad, if the borrower's CreditScore is below 700.
If a borrower has a high ProsperScore of 11.0 and a high CreditScoreRangeLower above 700, then his/her loan is almost guaranteed to not go bad.
Median and low ProsperScore (< 9.0) in the Term 36 coontributed most to bad loans.

Surprisingly, even if the `ProsperScore` was recorded as high as 10.0, it is still possible for the loans to go bad. Meanwhile, high `Credit Score` alone was not sufficient enough to guarantee the sfaety of loans.¶

	Term	BorrowerAPR	BorrowerRate	ProsperScore	CreditScoreRangeLower	CreditScoreRangeUpper	StatedMonthlyIncome	PercentFunded	Recommendations	InvestmentFromFriendsCount
count	113937.000000	113912.000000	113937.000000	84853.000000	113346.000000	113346.000000	1.139370e+05	113937.000000	113937.000000	113937.000000
mean	40.830248	0.218828	0.192764	5.950067	685.567731	704.567731	5.608026e+03	0.998584	0.048027	0.023460
std	10.436212	0.080364	0.074818	2.376501	66.458275	66.458275	7.478497e+03	0.017919	0.332353	0.232412
min	12.000000	0.006530	0.000000	1.000000	0.000000	19.000000	0.000000e+00	0.700000	0.000000	0.000000
25%	36.000000	0.156290	0.134000	4.000000	660.000000	679.000000	3.200333e+03	1.000000	0.000000	0.000000
50%	36.000000	0.209760	0.184000	6.000000	680.000000	699.000000	4.666667e+03	1.000000	0.000000	0.000000
75%	36.000000	0.283810	0.250000	8.000000	720.000000	739.000000	6.825000e+03	1.000000	0.000000	0.000000
max	60.000000	0.512290	0.497500	11.000000	880.000000	899.000000	1.750003e+06	1.012500	39.000000	33.000000

	Term	BorrowerAPR	BorrowerRate	ProsperScore	CreditScoreRangeLower	CreditScoreRangeUpper	StatedMonthlyIncome	PercentFunded	Recommendations	InvestmentFromFriendsCount
count	113937.000000	113912.000000	113937.000000	84853.000000	113346.000000	113346.000000	1.139370e+05	113937.000000	113937.000000	113937.000000
mean	40.830248	0.218828	0.192764	5.950067	685.567731	704.567731	5.608026e+03	0.998584	0.048027	0.023460
std	10.436212	0.080364	0.074818	2.376501	66.458275	66.458275	7.478497e+03	0.017919	0.332353	0.232412
min	12.000000	0.006530	0.000000	1.000000	0.000000	19.000000	0.000000e+00	0.700000	0.000000	0.000000
25%	36.000000	0.156290	0.134000	4.000000	660.000000	679.000000	3.200333e+03	1.000000	0.000000	0.000000
50%	36.000000	0.209760	0.184000	6.000000	680.000000	699.000000	4.666667e+03	1.000000	0.000000	0.000000
75%	36.000000	0.283810	0.250000	8.000000	720.000000	739.000000	6.825000e+03	1.000000	0.000000	0.000000
max	60.000000	0.512290	0.497500	11.000000	880.000000	899.000000	1.750003e+06	1.012500	39.000000	33.000000

	Term	LoanStatus	BorrowerAPR	BorrowerRate	ProsperScore	EmploymentStatus	IsBorrowerHomeowner	CreditScoreRangeLower	CreditScoreRangeUpper	StatedMonthlyIncome	PercentFunded
3274	36	Completed	0.06576	0.0590	NaN	Full-time	True	780.0	799.0	5.000000e+04	1.0000
4691	36	Completed	0.12528	0.0974	8.0	Employed	True	700.0	719.0	7.500000e+04	1.0000
8066	36	Completed	0.35643	0.3199	4.0	Employed	True	660.0	679.0	4.166667e+05	1.0000
8870	36	Chargedoff	0.12700	0.1200	NaN	Not available	True	740.0	759.0	2.083333e+05	1.0000
11312	36	Completed	0.08684	0.0800	NaN	Full-time	False	780.0	799.0	7.500000e+04	1.0000
11835	36	Current	0.13697	0.1089	9.0	Employed	True	720.0	739.0	5.250000e+04	1.0000
13039	36	Current	0.19236	0.1559	4.0	Self-employed	True	780.0	799.0	8.333333e+04	1.0000
15035	36	Current	0.14714	0.1189	10.0	Employed	True	720.0	739.0	6.666667e+04	1.0000
15355	36	Completed	0.18268	0.1645	NaN	Retired	True	700.0	719.0	8.333333e+04	1.0000
15814	36	Defaulted	0.21739	0.2100	NaN	Full-time	True	700.0	719.0	5.861667e+04	1.0000
17411	36	Completed	0.35643	0.3199	5.0	Employed	True	660.0	679.0	4.833333e+05	1.0000
17598	36	Completed	0.08684	0.0800	NaN	Not available	False	720.0	739.0	6.100000e+04	1.0000
20708	36	Completed	0.24758	0.2099	7.0	Employed	True	700.0	719.0	7.836442e+04	1.0000
21228	36	Current	0.32446	0.2850	2.0	Employed	True	640.0	659.0	5.333333e+04	1.0000
21250	36	Completed	0.16056	0.1390	10.0	Employed	True	760.0	779.0	5.333333e+04	0.7331
21983	60	Current	0.13227	0.1099	8.0	Self-employed	True	780.0	799.0	5.992742e+04	1.0000
25495	36	Completed	0.20614	0.1988	NaN	Full-time	False	600.0	619.0	6.499467e+04	1.0000
25803	36	Current	0.28032	0.2419	4.0	Self-employed	False	640.0	659.0	5.833333e+04	1.0000
27013	36	Completed	0.20200	0.1799	8.0	Employed	True	740.0	759.0	7.500000e+04	1.0000
28187	36	Current	0.21290	0.1760	4.0	Self-employed	False	680.0	699.0	7.500000e+04	1.0000
28298	60	Current	0.20347	0.1795	4.0	Self-employed	False	700.0	719.0	7.083333e+04	1.0000
29348	12	Completed	0.17493	0.1177	10.0	Employed	True	700.0	719.0	7.333333e+04	1.0000
31388	60	Current	0.20040	0.1765	3.0	Self-employed	True	700.0	719.0	5.000000e+04	1.0000
32104	12	Completed	0.18829	0.1299	8.0	Employed	True	820.0	839.0	5.000000e+04	1.0000
34950	60	Current	0.21935	0.1950	4.0	Employed	True	680.0	699.0	6.250000e+04	1.0000
35218	36	Completed	0.07389	0.0705	10.0	Full-time	True	860.0	879.0	1.083333e+05	1.0000
35833	36	Current	0.18275	0.1465	10.0	Employed	False	640.0	659.0	7.500000e+04	1.0000
38795	36	Current	0.14162	0.1203	7.0	Employed	True	780.0	799.0	5.191533e+04	1.0000
40276	36	Completed	0.14448	0.1374	NaN	Full-time	True	780.0	799.0	5.000000e+04	1.0000
44204	36	Current	0.16324	0.1274	7.0	Employed	True	700.0	719.0	7.444892e+04	1.0000
44367	36	Current	0.16259	0.1264	4.0	Self-employed	True	720.0	739.0	6.250000e+04	1.0000
44759	60	Current	0.20808	0.1840	4.0	Employed	True	720.0	739.0	6.666667e+04	1.0000
45028	36	Current	0.33973	0.2999	2.0	Self-employed	False	620.0	639.0	1.500000e+05	1.0000
45563	36	Current	0.21342	0.1765	4.0	Self-employed	False	700.0	719.0	7.500000e+04	1.0000
45648	36	Completed	0.09688	0.0900	NaN	Full-time	True	760.0	779.0	7.591667e+04	1.0000
48469	36	Completed	0.35643	0.3199	5.0	Self-employed	False	660.0	679.0	1.000000e+05	1.0000
49248	36	Completed	0.13068	0.1095	9.0	Employed	True	820.0	839.0	5.000000e+04	1.0000
49929	36	Current	0.35797	0.3177	3.0	Employed	True	700.0	719.0	5.000000e+04	1.0000
50126	36	Chargedoff	0.34577	0.3058	8.0	Employed	False	640.0	659.0	6.000000e+04	1.0000
51553	36	Current	0.09434	0.0809	11.0	Employed	True	700.0	719.0	7.666667e+04	1.0000
52293	36	Current	0.12782	0.0999	10.0	Employed	False	720.0	739.0	8.333333e+04	1.0000
53014	12	Completed	0.25785	0.1795	3.0	Self-employed	True	720.0	739.0	6.185478e+05	1.0000
53167	36	Current	0.35797	0.3177	4.0	Self-employed	True	740.0	759.0	1.750003e+06	1.0000
53386	36	Current	0.18725	0.1509	5.0	Self-employed	False	800.0	819.0	6.666667e+04	1.0000
56431	36	Completed	0.06587	0.0625	10.0	Employed	False	780.0	799.0	5.000000e+04	1.0000
56867	36	Completed	0.11144	0.1045	NaN	Self-employed	True	820.0	839.0	1.850818e+05	1.0000
57133	36	Current	0.35643	0.3199	5.0	Employed	True	680.0	699.0	4.666667e+05	1.0000
59263	36	Completed	0.27060	0.2324	5.0	Self-employed	True	800.0	819.0	8.333333e+04	1.0000
59298	36	Current	0.24246	0.2049	6.0	Self-employed	False	760.0	779.0	1.208333e+05	1.0000
59542	60	Completed	0.24282	0.2179	4.0	Employed	True	680.0	699.0	7.836442e+04	1.0000
59928	36	Completed	0.22135	0.1990	9.0	Full-time	False	680.0	699.0	9.626650e+04	1.0000
60563	36	Current	0.18173	0.1599	9.0	Employed	True	700.0	719.0	7.916667e+04	1.0000
61443	36	Completed	0.10692	0.1000	NaN	Full-time	True	800.0	819.0	6.666667e+04	1.0000
62274	36	Completed	0.17260	0.1654	NaN	Full-time	True	740.0	759.0	6.079108e+04	1.0000
62864	36	Defaulted	0.15211	0.1450	NaN	Self-employed	True	720.0	739.0	5.133333e+04	1.0000
63149	36	Current	0.14857	0.1203	7.0	Employed	True	860.0	879.0	9.166667e+04	1.0000
63374	36	Completed	0.31053	0.2750	7.0	Employed	True	700.0	719.0	7.083333e+04	1.0000
63925	36	Completed	0.25948	0.2500	NaN	Not available	True	560.0	579.0	1.404167e+05	1.0000
65042	36	Chargedoff	0.15713	0.1500	NaN	Full-time	True	720.0	739.0	5.416667e+04	1.0000
65099	36	Current	0.09484	0.0814	10.0	Employed	True	820.0	839.0	7.666667e+04	1.0000
67643	36	Completed	0.08885	0.0820	NaN	Full-time	True	780.0	799.0	6.666667e+04	1.0000
67756	36	Current	0.21372	0.1768	8.0	Employed	True	720.0	739.0	6.250000e+04	1.0000
68703	36	Current	0.16324	0.1274	4.0	Employed	True	700.0	719.0	7.444892e+04	1.0000
69108	36	Completed	0.22744	0.2200	NaN	Not available	False	560.0	579.0	8.000000e+04	1.0000
72473	36	Current	0.20268	0.1660	3.0	Employed	True	700.0	719.0	5.000000e+04	1.0000
75295	36	Current	0.22108	0.1840	6.0	Employed	False	680.0	699.0	5.500000e+04	1.0000
76102	36	Completed	0.22237	0.2000	5.0	Self-employed	True	780.0	799.0	7.166667e+04	1.0000
76865	36	Current	0.35797	0.3177	4.0	Employed	True	680.0	699.0	7.500000e+04	1.0000
77941	36	Current	0.35797	0.3177	2.0	Self-employed	True	760.0	779.0	7.083333e+04	1.0000
78431	36	Chargedoff	0.20253	0.1952	NaN	Full-time	False	640.0	659.0	1.033341e+05	1.0000
80075	36	Current	0.20217	0.1655	6.0	Employed	True	700.0	719.0	1.583333e+05	1.0000
80092	12	Completed	0.18526	0.1279	8.0	Employed	False	680.0	699.0	6.666667e+04	1.0000
80516	12	Current	0.25785	0.1795	2.0	Self-employed	True	760.0	779.0	5.416667e+04	1.0000
82831	36	Current	0.31012	0.2710	3.0	Self-employed	True	700.0	719.0	2.500000e+05	1.0000
83184	36	Completed	0.09223	0.0750	NaN	Not available	True	840.0	859.0	7.166667e+04	1.0000
83697	36	Completed	0.08684	0.0800	NaN	Not available	True	840.0	859.0	1.850818e+05	1.0000
89928	36	Defaulted	0.16014	0.1530	NaN	Full-time	True	660.0	679.0	5.250000e+04	1.0000
90539	60	Current	0.20436	0.1899	8.0	Employed	True	700.0	719.0	6.666667e+04	0.7328
91942	36	Chargedoff	0.25205	0.2445	NaN	Not available	True	700.0	719.0	5.227833e+04	1.0000
93680	36	Completed	0.12500	0.1180	NaN	Full-time	False	680.0	699.0	7.833333e+04	1.0000
93996	60	Current	0.20347	0.1795	10.0	Self-employed	False	700.0	719.0	7.083333e+04	1.0000
100005	36	Current	0.12691	0.0990	3.0	Self-employed	True	760.0	779.0	5.500000e+04	1.0000
100431	36	Chargedoff	0.29776	0.2900	NaN	Full-time	True	640.0	659.0	7.035000e+04	1.0000
103950	60	Current	0.12211	0.1199	10.0	Employed	True	760.0	779.0	5.416667e+04	0.7027
104608	60	Current	0.18555	0.1620	8.0	Employed	True	680.0	699.0	6.666667e+04	1.0000
104834	36	Current	0.20564	0.1764	8.0	Employed	True	720.0	739.0	7.083333e+04	1.0000
105248	36	Current	0.20562	0.1685	5.0	Employed	True	700.0	719.0	7.083333e+04	1.0000
108154	36	Current	0.14857	0.1203	9.0	Employed	True	720.0	739.0	5.250000e+04	1.0000
109043	36	Current	0.18214	0.1459	7.0	Employed	True	700.0	719.0	6.875000e+04	1.0000
111265	60	Completed	0.20593	0.1819	9.0	Employed	False	640.0	659.0	1.087500e+05	1.0000
113270	36	Current	0.22712	0.1899	6.0	Employed	False	680.0	699.0	3.944000e+05	1.0000

	Term	LoanStatus	BorrowerAPR	BorrowerRate	ProsperScore	EmploymentStatus	IsBorrowerHomeowner	CreditScoreRangeLower	CreditScoreRangeUpper	StatedMonthlyIncome	PercentFunded
3274	36	Completed	0.06576	0.0590	NaN	Full-time	True	780.0	799.0	5.000000e+04	1.0000
4691	36	Completed	0.12528	0.0974	8.0	Employed	True	700.0	719.0	7.500000e+04	1.0000
8066	36	Completed	0.35643	0.3199	4.0	Employed	True	660.0	679.0	4.166667e+05	1.0000
8870	36	Chargedoff	0.12700	0.1200	NaN	Not available	True	740.0	759.0	2.083333e+05	1.0000
11312	36	Completed	0.08684	0.0800	NaN	Full-time	False	780.0	799.0	7.500000e+04	1.0000
11835	36	Current	0.13697	0.1089	9.0	Employed	True	720.0	739.0	5.250000e+04	1.0000
13039	36	Current	0.19236	0.1559	4.0	Self-employed	True	780.0	799.0	8.333333e+04	1.0000
15035	36	Current	0.14714	0.1189	10.0	Employed	True	720.0	739.0	6.666667e+04	1.0000
15355	36	Completed	0.18268	0.1645	NaN	Retired	True	700.0	719.0	8.333333e+04	1.0000
15814	36	Defaulted	0.21739	0.2100	NaN	Full-time	True	700.0	719.0	5.861667e+04	1.0000
17411	36	Completed	0.35643	0.3199	5.0	Employed	True	660.0	679.0	4.833333e+05	1.0000
17598	36	Completed	0.08684	0.0800	NaN	Not available	False	720.0	739.0	6.100000e+04	1.0000
20708	36	Completed	0.24758	0.2099	7.0	Employed	True	700.0	719.0	7.836442e+04	1.0000
21228	36	Current	0.32446	0.2850	2.0	Employed	True	640.0	659.0	5.333333e+04	1.0000
21250	36	Completed	0.16056	0.1390	10.0	Employed	True	760.0	779.0	5.333333e+04	0.7331
21983	60	Current	0.13227	0.1099	8.0	Self-employed	True	780.0	799.0	5.992742e+04	1.0000
25495	36	Completed	0.20614	0.1988	NaN	Full-time	False	600.0	619.0	6.499467e+04	1.0000
25803	36	Current	0.28032	0.2419	4.0	Self-employed	False	640.0	659.0	5.833333e+04	1.0000
27013	36	Completed	0.20200	0.1799	8.0	Employed	True	740.0	759.0	7.500000e+04	1.0000
28187	36	Current	0.21290	0.1760	4.0	Self-employed	False	680.0	699.0	7.500000e+04	1.0000
28298	60	Current	0.20347	0.1795	4.0	Self-employed	False	700.0	719.0	7.083333e+04	1.0000
29348	12	Completed	0.17493	0.1177	10.0	Employed	True	700.0	719.0	7.333333e+04	1.0000
31388	60	Current	0.20040	0.1765	3.0	Self-employed	True	700.0	719.0	5.000000e+04	1.0000
32104	12	Completed	0.18829	0.1299	8.0	Employed	True	820.0	839.0	5.000000e+04	1.0000
34950	60	Current	0.21935	0.1950	4.0	Employed	True	680.0	699.0	6.250000e+04	1.0000
35218	36	Completed	0.07389	0.0705	10.0	Full-time	True	860.0	879.0	1.083333e+05	1.0000
35833	36	Current	0.18275	0.1465	10.0	Employed	False	640.0	659.0	7.500000e+04	1.0000
38795	36	Current	0.14162	0.1203	7.0	Employed	True	780.0	799.0	5.191533e+04	1.0000
40276	36	Completed	0.14448	0.1374	NaN	Full-time	True	780.0	799.0	5.000000e+04	1.0000
44204	36	Current	0.16324	0.1274	7.0	Employed	True	700.0	719.0	7.444892e+04	1.0000
44367	36	Current	0.16259	0.1264	4.0	Self-employed	True	720.0	739.0	6.250000e+04	1.0000
44759	60	Current	0.20808	0.1840	4.0	Employed	True	720.0	739.0	6.666667e+04	1.0000
45028	36	Current	0.33973	0.2999	2.0	Self-employed	False	620.0	639.0	1.500000e+05	1.0000
45563	36	Current	0.21342	0.1765	4.0	Self-employed	False	700.0	719.0	7.500000e+04	1.0000
45648	36	Completed	0.09688	0.0900	NaN	Full-time	True	760.0	779.0	7.591667e+04	1.0000
48469	36	Completed	0.35643	0.3199	5.0	Self-employed	False	660.0	679.0	1.000000e+05	1.0000
49248	36	Completed	0.13068	0.1095	9.0	Employed	True	820.0	839.0	5.000000e+04	1.0000
49929	36	Current	0.35797	0.3177	3.0	Employed	True	700.0	719.0	5.000000e+04	1.0000
50126	36	Chargedoff	0.34577	0.3058	8.0	Employed	False	640.0	659.0	6.000000e+04	1.0000
51553	36	Current	0.09434	0.0809	11.0	Employed	True	700.0	719.0	7.666667e+04	1.0000
52293	36	Current	0.12782	0.0999	10.0	Employed	False	720.0	739.0	8.333333e+04	1.0000
53014	12	Completed	0.25785	0.1795	3.0	Self-employed	True	720.0	739.0	6.185478e+05	1.0000
53167	36	Current	0.35797	0.3177	4.0	Self-employed	True	740.0	759.0	1.750003e+06	1.0000
53386	36	Current	0.18725	0.1509	5.0	Self-employed	False	800.0	819.0	6.666667e+04	1.0000
56431	36	Completed	0.06587	0.0625	10.0	Employed	False	780.0	799.0	5.000000e+04	1.0000
56867	36	Completed	0.11144	0.1045	NaN	Self-employed	True	820.0	839.0	1.850818e+05	1.0000
57133	36	Current	0.35643	0.3199	5.0	Employed	True	680.0	699.0	4.666667e+05	1.0000
59263	36	Completed	0.27060	0.2324	5.0	Self-employed	True	800.0	819.0	8.333333e+04	1.0000
59298	36	Current	0.24246	0.2049	6.0	Self-employed	False	760.0	779.0	1.208333e+05	1.0000
59542	60	Completed	0.24282	0.2179	4.0	Employed	True	680.0	699.0	7.836442e+04	1.0000
59928	36	Completed	0.22135	0.1990	9.0	Full-time	False	680.0	699.0	9.626650e+04	1.0000
60563	36	Current	0.18173	0.1599	9.0	Employed	True	700.0	719.0	7.916667e+04	1.0000
61443	36	Completed	0.10692	0.1000	NaN	Full-time	True	800.0	819.0	6.666667e+04	1.0000
62274	36	Completed	0.17260	0.1654	NaN	Full-time	True	740.0	759.0	6.079108e+04	1.0000
62864	36	Defaulted	0.15211	0.1450	NaN	Self-employed	True	720.0	739.0	5.133333e+04	1.0000
63149	36	Current	0.14857	0.1203	7.0	Employed	True	860.0	879.0	9.166667e+04	1.0000
63374	36	Completed	0.31053	0.2750	7.0	Employed	True	700.0	719.0	7.083333e+04	1.0000
63925	36	Completed	0.25948	0.2500	NaN	Not available	True	560.0	579.0	1.404167e+05	1.0000
65042	36	Chargedoff	0.15713	0.1500	NaN	Full-time	True	720.0	739.0	5.416667e+04	1.0000
65099	36	Current	0.09484	0.0814	10.0	Employed	True	820.0	839.0	7.666667e+04	1.0000
67643	36	Completed	0.08885	0.0820	NaN	Full-time	True	780.0	799.0	6.666667e+04	1.0000
67756	36	Current	0.21372	0.1768	8.0	Employed	True	720.0	739.0	6.250000e+04	1.0000
68703	36	Current	0.16324	0.1274	4.0	Employed	True	700.0	719.0	7.444892e+04	1.0000
69108	36	Completed	0.22744	0.2200	NaN	Not available	False	560.0	579.0	8.000000e+04	1.0000
72473	36	Current	0.20268	0.1660	3.0	Employed	True	700.0	719.0	5.000000e+04	1.0000
75295	36	Current	0.22108	0.1840	6.0	Employed	False	680.0	699.0	5.500000e+04	1.0000
76102	36	Completed	0.22237	0.2000	5.0	Self-employed	True	780.0	799.0	7.166667e+04	1.0000
76865	36	Current	0.35797	0.3177	4.0	Employed	True	680.0	699.0	7.500000e+04	1.0000
77941	36	Current	0.35797	0.3177	2.0	Self-employed	True	760.0	779.0	7.083333e+04	1.0000
78431	36	Chargedoff	0.20253	0.1952	NaN	Full-time	False	640.0	659.0	1.033341e+05	1.0000
80075	36	Current	0.20217	0.1655	6.0	Employed	True	700.0	719.0	1.583333e+05	1.0000
80092	12	Completed	0.18526	0.1279	8.0	Employed	False	680.0	699.0	6.666667e+04	1.0000
80516	12	Current	0.25785	0.1795	2.0	Self-employed	True	760.0	779.0	5.416667e+04	1.0000
82831	36	Current	0.31012	0.2710	3.0	Self-employed	True	700.0	719.0	2.500000e+05	1.0000
83184	36	Completed	0.09223	0.0750	NaN	Not available	True	840.0	859.0	7.166667e+04	1.0000
83697	36	Completed	0.08684	0.0800	NaN	Not available	True	840.0	859.0	1.850818e+05	1.0000
89928	36	Defaulted	0.16014	0.1530	NaN	Full-time	True	660.0	679.0	5.250000e+04	1.0000
90539	60	Current	0.20436	0.1899	8.0	Employed	True	700.0	719.0	6.666667e+04	0.7328
91942	36	Chargedoff	0.25205	0.2445	NaN	Not available	True	700.0	719.0	5.227833e+04	1.0000
93680	36	Completed	0.12500	0.1180	NaN	Full-time	False	680.0	699.0	7.833333e+04	1.0000
93996	60	Current	0.20347	0.1795	10.0	Self-employed	False	700.0	719.0	7.083333e+04	1.0000
100005	36	Current	0.12691	0.0990	3.0	Self-employed	True	760.0	779.0	5.500000e+04	1.0000
100431	36	Chargedoff	0.29776	0.2900	NaN	Full-time	True	640.0	659.0	7.035000e+04	1.0000
103950	60	Current	0.12211	0.1199	10.0	Employed	True	760.0	779.0	5.416667e+04	0.7027
104608	60	Current	0.18555	0.1620	8.0	Employed	True	680.0	699.0	6.666667e+04	1.0000
104834	36	Current	0.20564	0.1764	8.0	Employed	True	720.0	739.0	7.083333e+04	1.0000
105248	36	Current	0.20562	0.1685	5.0	Employed	True	700.0	719.0	7.083333e+04	1.0000
108154	36	Current	0.14857	0.1203	9.0	Employed	True	720.0	739.0	5.250000e+04	1.0000
109043	36	Current	0.18214	0.1459	7.0	Employed	True	700.0	719.0	6.875000e+04	1.0000
111265	60	Completed	0.20593	0.1819	9.0	Employed	False	640.0	659.0	1.087500e+05	1.0000
113270	36	Current	0.22712	0.1899	6.0	Employed	False	680.0	699.0	3.944000e+05	1.0000

	ListingKey	ListingNumber	ListingCreationDate	CreditGrade	Term	LoanStatus	ClosedDate	BorrowerAPR	BorrowerRate	LenderYield	...	LP_ServiceFees	PercentFunded	Investors
0	1021339766868145413AB3B	193129	2007-08-26 19:09:29.263000000	C	36	Completed	2009-08-14 00:00:00	0.16516	0.1580	0.1380	...	-133.18	1.0	258
1	10273602499503308B223C1	1209647	2014-02-27 08:28:07.900000000	NaN	36	Current	NaN	0.12016	0.0920	0.0820	...	0.00	1.0	1
2	0EE9337825851032864889A	81716	2007-01-05 15:00:47.090000000	HR	36	Completed	2009-12-17 00:00:00	0.28269	0.2750	0.2400	...	-24.20	1.0	41
3	0EF5356002482715299901A	658116	2012-10-22 11:02:35.010000000	NaN	36	Current	NaN	0.12528	0.0974	0.0874	...	-108.01	1.0	158
4	0F023589499656230C5E3E2	909464	2013-09-14 18:38:39.097000000	NaN	36	Current	NaN	0.24614	0.2085	0.1985	...	-60.27	1.0	20

	Variable	Description
0	ListingKey	Unique key for each listing, same value as the 'key' used in the listing object in the API.
1	ListingNumber	The number that uniquely identifies the listing to the public as displayed on the website.
2	ListingCreationDate	The date the listing was created.
3	CreditGrade	The Credit rating that was assigned at the time the listing went live. Applicable for listings pre-2009 period and will only be populated for those listings.
4	Term	The length of the loan expressed in months.
5	LoanStatus	The current status of the loan: Cancelled, Chargedoff, Completed, Current, Defaulted, FinalPaymentInProgress, PastDue. The PastDue status will be accompanied by a delinquency bucket.
6	ClosedDate	Closed date is applicable for Cancelled, Completed, Chargedoff and Defaulted loan statuses.
7	BorrowerAPR	The Borrower's Annual Percentage Rate (APR) for the loan.
8	BorrowerRate	The Borrower's interest rate for this loan.
9	LenderYield	The Lender yield on the loan. Lender yield is equal to the interest rate on the loan less the servicing fee.
10	EstimatedEffectiveYield	Effective yield is equal to the borrower interest rate (i) minus the servicing fee rate, (ii) minus estimated uncollected interest on charge-offs, (iii) plus estimated collected late fees. Applicable for loans originated after July 2009.
11	EstimatedLoss	Estimated loss is the estimated principal loss on charge-offs. Applicable for loans originated after July 2009.
12	EstimatedReturn	The estimated return assigned to the listing at the time it was created. Estimated return is the difference between the Estimated Effective Yield and the Estimated Loss Rate. Applicable for loans originated after July 2009.
13	ProsperRating (numeric)	The Prosper Rating assigned at the time the listing was created: 0 - N/A, 1 - HR, 2 - E, 3 - D, 4 - C, 5 - B, 6 - A, 7 - AA. Applicable for loans originated after July 2009.
14	ProsperRating (Alpha)	The Prosper Rating assigned at the time the listing was created between AA - HR. Applicable for loans originated after July 2009.
15	ProsperScore	A custom risk score built using historical Prosper data. The score ranges from 1-10, with 10 being the best, or lowest risk score. Applicable for loans originated after July 2009.
16	ListingCategory	The category of the listing that the borrower selected when posting their listing: 0 - Not Available, 1 - Debt Consolidation, 2 - Home Improvement, 3 - Business, 4 - Personal Loan, 5 - Student Use, 6 - Auto, 7- Other, 8 - Baby&Adoption, 9 - Boat, 10 - Cosmetic Procedure, 11 - Engagement Ring, 12 - Green Loans, 13 - Household Expenses, 14 - Large Purchases, 15 - Medical/Dental, 16 - Motorcycle, 17 - RV, 18 - Taxes, 19 - Vacation, 20 - Wedding Loans
17	BorrowerState	The two letter abbreviation of the state of the address of the borrower at the time the Listing was created.
18	Occupation	The Occupation selected by the Borrower at the time they created the listing.
19	EmploymentStatus	The employment status of the borrower at the time they posted the listing.
20	EmploymentStatusDuration	The length in months of the employment status at the time the listing was created.
21	IsBorrowerHomeowner	A Borrower will be classified as a homowner if they have a mortgage on their credit profile or provide documentation confirming they are a homeowner.
22	CurrentlyInGroup	Specifies whether or not the Borrower was in a group at the time the listing was created.
23	GroupKey	The Key of the group in which the Borrower is a member of. Value will be null if the borrower does not have a group affiliation.
24	DateCreditPulled	The date the credit profile was pulled.
25	CreditScoreRangeLower	The lower value representing the range of the borrower's credit score as provided by a consumer credit rating agency.
26	CreditScoreRangeUpper	The upper value representing the range of the borrower's credit score as provided by a consumer credit rating agency.
27	FirstRecordedCreditLine	The date the first credit line was opened.
28	CurrentCreditLines	Number of current credit lines at the time the credit profile was pulled.
29	OpenCreditLines	Number of open credit lines at the time the credit profile was pulled.
30	TotalCreditLinespast7years	Number of credit lines in the past seven years at the time the credit profile was pulled.
31	OpenRevolvingAccounts	Number of open revolving accounts at the time the credit profile was pulled.
32	OpenRevolvingMonthlyPayment	Monthly payment on revolving accounts at the time the credit profile was pulled.
33	InquiriesLast6Months	Number of inquiries in the past six months at the time the credit profile was pulled.
34	TotalInquiries	Total number of inquiries at the time the credit profile was pulled.
35	CurrentDelinquencies	Number of accounts delinquent at the time the credit profile was pulled.
36	AmountDelinquent	Dollars delinquent at the time the credit profile was pulled.
37	DelinquenciesLast7Years	Number of delinquencies in the past 7 years at the time the credit profile was pulled.
38	PublicRecordsLast10Years	Number of public records in the past 10 years at the time the credit profile was pulled.
39	PublicRecordsLast12Months	Number of public records in the past 12 months at the time the credit profile was pulled.
40	RevolvingCreditBalance	Dollars of revolving credit at the time the credit profile was pulled.
41	BankcardUtilization	The percentage of available revolving credit that is utilized at the time the credit profile was pulled.
42	AvailableBankcardCredit	The total available credit via bank card at the time the credit profile was pulled.
43	TotalTrades	Number of trade lines ever opened at the time the credit profile was pulled.
44	TradesNeverDelinquent	Number of trades that have never been delinquent at the time the credit profile was pulled.
45	TradesOpenedLast6Months	Number of trades opened in the last 6 months at the time the credit profile was pulled.
46	DebtToIncomeRatio	The debt to income ratio of the borrower at the time the credit profile was pulled. This value is Null if the debt to income ratio is not available. This value is capped at 10.01 (any debt to income ratio larger than 1000% will be returned as 1001%).
47	IncomeRange	The income range of the borrower at the time the listing was created.
48	IncomeVerifiable	The borrower indicated they have the required documentation to support their income.
49	StatedMonthlyIncome	The monthly income the borrower stated at the time the listing was created.
50	LoanKey	Unique key for each loan. This is the same key that is used in the API.
51	TotalProsperLoans	Number of Prosper loans the borrower at the time they created this listing. This value will be null if the borrower had no prior loans.
52	TotalProsperPaymentsBilled	Number of on time payments the borrower made on Prosper loans at the time they created this listing. This value will be null if the borrower had no prior loans.
53	OnTimeProsperPayments	Number of on time payments the borrower had made on Prosper loans at the time they created this listing. This value will be null if the borrower has no prior loans.
54	ProsperPaymentsLessThanOneMonthLate	Number of payments the borrower made on Prosper loans that were less than one month late at the time they created this listing. This value will be null if the borrower had no prior loans.
55	ProsperPaymentsOneMonthPlusLate	Number of payments the borrower made on Prosper loans that were greater than one month late at the time they created this listing. This value will be null if the borrower had no prior loans.
56	ProsperPrincipalBorrowed	Total principal borrowed on Prosper loans at the time the listing was created. This value will be null if the borrower had no prior loans.
57	ProsperPrincipalOutstanding	Principal outstanding on Prosper loans at the time the listing was created. This value will be null if the borrower had no prior loans.
58	ScorexChangeAtTimeOfListing	Borrower's credit score change at the time the credit profile was pulled. This will be the change relative to the borrower's last Prosper loan. This value will be null if the borrower had no prior loans.
59	LoanCurrentDaysDelinquent	The number of days delinquent.
60	LoanFirstDefaultedCycleNumber	The cycle the loan was charged off. If the loan has not charged off the value will be null.
61	LoanMonthsSinceOrigination	Number of months since the loan originated.
62	LoanNumber	Unique numeric value associated with the loan.
63	LoanOriginalAmount	The origination amount of the loan.
64	LoanOriginationDate	The date the loan was originated.
65	LoanOriginationQuarter	The quarter in which the loan was originated.
66	MemberKey	The unique key that is associated with the borrower. This is the same identifier that is used in the API member object.
67	MonthlyLoanPayment	The scheduled monthly loan payment.
68	LP_CustomerPayments	Pre charge-off cumulative gross payments made by the borrower on the loan. If the loan has charged off, this value will exclude any recoveries.
69	LP_CustomerPrincipalPayments	Pre charge-off cumulative principal payments made by the borrower on the loan. If the loan has charged off, this value will exclude any recoveries.
70	LP_InterestandFees	Pre charge-off cumulative interest and fees paid by the borrower. If the loan has charged off, this value will exclude any recoveries.
71	LP_ServiceFees	Cumulative service fees paid by the investors who have invested in the loan.
72	LP_CollectionFees	Cumulative collection fees paid by the investors who have invested in the loan.
73	LP_GrossPrincipalLoss	The gross charged off amount of the loan.
74	LP_NetPrincipalLoss	The principal that remains uncollected after any recoveries.
75	LP_NonPrincipalRecoverypayments	The interest and fee component of any recovery payments. The current payment policy applies payments in the following order: Fees, interest, principal.
76	PercentFunded	Percent the listing was funded.
77	Recommendations	Number of recommendations the borrower had at the time the listing was created.
78	InvestmentFromFriendsCount	Number of friends that made an investment in the loan.
79	InvestmentFromFriendsAmount	Dollar amount of investments that were made by friends.
80	Investors	The number of investors that funded the loan.

Loan Data From Prosper¶

by Peter Wu¶

Preliminary Wrangling¶

I am interested in two things: first, which features are best at predicting the LoanStatus; second, which features determine the BorrowerAPR and BorrowerRate at most?¶

I expect the two CreditScoreRange will have the most effect on loan completion, and variables like Occupation, EmploymentStatus, IsBorrowerHomeowner will determine the borrowers's interest rate.¶

Univariate Exploration¶

The StatedMonthlyIncome took a very long range of values, as such a log transform was applied. After the transformation ,the variable looks to be normally distributed, but still with a extremely long tail in both directions.¶

During the above process, many invalid income values were found, and were excluded from the dataset, a significant amount of outliers were thus removed. This is important for bivariate and multivariate analysis in the later part, because those values would enormously drag the average income low.¶

Briefly, after the univariate exploration, three variables were deemed to be less value of analysis, and would not be used in the next part. They are:¶

PercentFunded, Recommendations, InvestmentFromFriendsCount.¶

The rest will be used to move forwards are:¶

Term, LoanStatus, BorrowerAPR, BorrowerRate, EmploymentStatus, IsBorrowerHomeowner,¶

CreditScoreRangeLower, ProsperScore, CreditScoreRangeUpper, and StatedMonthlyIncome (10 variables in total).¶

Bivariate Exploration¶

This part will focus on the relationship between: first, BorrowerAPR and BorrowerRate and the other variables, to dig out the¶

determinants of brrowers' interest rate; second, LoanStatusand the other variables, to dig out what would be the potential reason for a loan to go bad.¶

I will start with BorrowerAPR and BorrowerRate (Borrowers rate).¶

The disciussion will now shift from BorrowerAPR and BorrowerRate to LoanStatus, to see how other variables interact with it.¶

For BorrowerAPR and BorrowerRate: the pair of CreditScore Bounds and ProsperScore seemed to be both negatively related to them.¶

For LoanStatus: CreditScore Bounds, Term, ProsperScore seemed to affect it in various ways.¶

Interestingly, the two pairs of variables: BorrowerAPR and BorrowerRate, CreditScoreRangeLower and CreditScoreRangeUpper are almost perfectly related to each other.¶

There is also an interplay betweeen the three pivot variables, although their relationships were not intended to be analyzed initially.¶